A Study on Real-world Effectiveness of Model Year 2015–2023 Advanced Driver Assistance Systems

Largest automatic emergency braking study finds systems improving over time.

Download Resources

Automatic emergency braking (AEB) continues to improve as it cuts rear-end crashes in half. The latest study by the Partnership for Analytics Research in Traffic Safety (PARTS), pairing auto manufacturer equipment with police crash report data covering 98 million vehicles and 21.2 million crashes, is the largest and most comprehensive study of advanced driver assistance systems (ADAS) to date.

This study is a follow-up to a previous ADAS effectiveness study released in 2022. It more than doubled the number of vehicle models included, and added three additional vehicle segments, three additional states, and three new model years. For the first time, the study included data from new PARTS members Ford and Hyundai. Other manufacturers contributing data to this study were General Motors, Honda, Mazda, Mitsubishi, Stellantis, Subaru, and Toyota.

The data showed an increase in AEB effectiveness, from 46% across model years 2015–2017 to 52% across model years 2021–2023, indicating that advancements in the technology have led to tangible improvements.

The data also showed a 9% reduction in single-vehicle frontal crashes with non-motorists, including pedestrians, cyclists, scooters, and wheelchairs, for vehicles equipped with pedestrian automatic emergency braking (PAEB) systems. This is the first time a statistically significant measure of PAEB effectiveness has been quantified by PARTS. Pedestrian crashes are among the most severe forms of traffic crashes, with deaths accounting for 18% of all traffic fatalities, according to NHTSA.

The data will be used to further study ADAS effectiveness in reducing crash severity. Future studies will incorporate data from Kia, the newest PARTS member. The effectiveness study considered whether a vehicle was equipped with a given ADAS feature at the time of manufacture, and future studies will attempt to assess whether that feature was on or activated at the time of crash.